Robust Bundle Adjustment for Structure from Motion

摘要: Structure from motion (SFM) is the problem of reconstructing geometry a scene stream images. In this problem, must be inferred images, along with camera pose parameters. Bundle adjustment (BA) refinement method used to improve SFM solutions. It consists in simultaneously improving set initial estimates for all parameters (structure and pose) by minimizing global cost function. generally considered highly accurate, so typically as last step most current methods. Unfortunately, estimating images an ill-conditioned problem. We thus propose BA formulation which does not involve solving orientations. tested approach on several real world models. The numerical results obtained show that much less affected noise than traditional BA.